fname='train_img.png'
trImg = cv.imread(fname, cv.IMREAD_GRAYSCALE);
ret, bwtrImg=cv.threshold(trImg, 122, 255, cv.THRESH_BINARY_INV)
bxs=extractBlobs(bwtrImg, False, True)
#bxs.sort()
# print(bxs)
samples = np.zeros((36, 69));
i=0
print(bxs[0])
for bx in bxs:
# print(trImg.shape)
cimg = trImg[bx[0]:bx[0]+bx[2],bx[1]:bx[1]+bx[3]];
# print(cimg.shape)
clbp = findLBP(cimg);
chist = lbpHist(clbp);
chist=chist.reshape((1, 69))
# print(chist.shape);
samples[i, :]=chist;
i+=1
# cv.imshow('img',cimg);
# cv.waitKey(0)
resp=['9','8','7','6','5','4','3','2','1','0','Z','Y','X','W','V','U','T','S','R','Q','P','O','N','H','F','E','D','B','M','L','K','J','C','A','I','G']
responses=[ord(i) for i in resp]
# print(resp)
responses=np.asarray(responses)
responses=responses.reshape((36,1))
# print(responses)
print(responses.shape)
print(samples.shape)
with open('samples.data','wb') as f:
import pickle
tImg = cv.imread('image_0040.jpg',cv.IMREAD_GRAYSCALE);
img = cv.imread('cars_markus/image_0011.jpg',cv.IMREAD_GRAYSCALE);
img=cv.GaussianBlur(img, (3,3), 0)
tImg=cv.GaussianBlur(tImg, (3,3), 0)
tImg = cv.Sobel(tImg,cv.CV_8U,1,0,ksize=3)
img = cv.Sobel(img,cv.CV_8U,1,0,ksize=3)
imLbp = findLBP(img);
tLbp = findLBP(tImg);
h,w = imLbp.shape;
tHist=lbpHist(tLbp);
tcHist=clrHist(tImg);
minDist=100000
bBox=[0,0,0,0];
scaleStp=15;
moveStp=10;
ratio=(tImg.shape[1]*1.0)/tImg.shape[0]
bxs=[]
for scaleF in range(1,scaleStp/3):
winW=int(scaleF*(w/scaleStp));
winH=int(winW/ratio);
print(scaleF);
print(minDist)
minDist=100000
